Tyler Cowen linked to an interview with Banerjee & Duflo a couple days ago, but I passed over it at the time, and only now that I’m going through my RSS feed have I decided to check it out. I just wanted to highlight a quote from part 1: “My interest in this question [what leads people to become entrepreneurs] came particularly from Scott Shane’s book The Illusions of Entrepreneurship where he points out that the best predictor of a person starting a business is whether they have a drug record. People with a drug record can’t get jobs so they start a business.”
If the name Scott Shane is familiar, it might be because I’ve linked this before.
UPDATE: Scott Shane replied that Tim Ogden got his claim wrong.
On a completely unrelated note, there has been discussion here on experiments in menu labeling and another study just came out, although if you try to go to the source only the abstract is in front of the paywall.
Entitled to an Opinion 
July 2, 2011 at 12:56 pm
I have to confess to being extremely skeptical of such a datapoint. Drug convictions correlate strongly with being poor or otherwise of low SES, and that makes starting businesses hard… I would expect IQ, wealth, education, or any of a dozen variables to be better predictors!
July 2, 2011 at 2:55 pm
America sustains the myth of entrepreneurship even though for most folks their attempt to be successful at it will fail. But people like Virginia Postrel defend this particular form of irrational exuberance:
http://www.bigquestionsonline.com/columns/virginia-postrel/in-praise-of-irrational-exuberance
No big surprise then that libertarians also defend the right to eat whatever you want, not be guided by technocrats in their retirement strategies (or lack thereof), etc. Their philosophy is generally supportive of the right of others to fuck up, even when you’re almost certain they will. But hey, if the journey is more important than the destination…
On the menu labels, you’ll notice the study’s designers concede information apparently won’t cut it, so you’ll just have to make more of the cafeteria food healthy so students can’t avoid eating better.
July 3, 2011 at 6:32 am
gwern beats me to it, but it sounds bullshitty.
But it’s also part of a larger genre of counterintuitive “single best predictor points” that sound bullshitty.
It reminds of the claim that IQ “is the single best predictor of job performance”. Does that have robust empirical backing or is it an overstated, de-nuanced argument?
July 3, 2011 at 4:12 pm
I have emailed Scott Shane asking if he has a citation for that claim.
One thing I like about Patri Friedman’s advocacy for “seasteads” is that he admits most will fail, and what eventually succeeds will be something nobody has yet thought up. With a large number of experiments, some will achieve the unlikely.
July 5, 2011 at 5:42 pm
[…] is a reply Scott Shane sent to me about Tim Ogden’s quote: Did someone really say that? That’s absurd. There is no way that having a drug record could […]
July 5, 2011 at 6:58 pm
TGGP is correct that the statement in the interview was an error. In fact, there are two different errors in the statement. Here’s the explanation I posted:
——————–
TGGP,
thanks for reading closely enough to catch my error. In fact there are two errors in my statement, with two different causes:
1) First I used the word “best” as a qualifier. In fact, as you note in your blog post, the source of Shane’s statistic in his book, a paper by Robert Fairlie (available here: http://bit.ly/pA9KH0 ) finds that dealing drugs as a youth is associated with an 11 to 21 percent increase in likelihood to start a business. That’s certainly a good predictor, far better of one than many people would assume, but by no means the “best”.
2) Second, upon seeing your comment I immediately went back and re-read the passage in question in Shane’s book. It turns out that cognitive biases intervened and I misread the passage in question.
Specifically, here’s the quotation from the book, p. 44: “Finally, people who dealt drugs as teenagers are between 11 and 21 percent more likely than other people to start their own businesses in adulthood. And their higher rate of self-employment isn’t the result of wealth accumulated dealing drugs, greater likelihood of having a criminal record, or lower wages.”
I read that second sentence as: And their higher rate of self-employment isn’t the result of wealth accumulated dealing drugs, [BUT] greater likelihood of having a criminal record, or lower wages.
My priors interfered with me absorbing what was really being said.
So thanks for helping me update those priors and catch an error in what I “know” that I never would have caught myself.